EP0703698A1 - Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect - Google Patents

Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect Download PDF

Info

Publication number
EP0703698A1
EP0703698A1 EP95117080A EP95117080A EP0703698A1 EP 0703698 A1 EP0703698 A1 EP 0703698A1 EP 95117080 A EP95117080 A EP 95117080A EP 95117080 A EP95117080 A EP 95117080A EP 0703698 A1 EP0703698 A1 EP 0703698A1
Authority
EP
European Patent Office
Prior art keywords
image
pixel
output
coordinate
input
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP95117080A
Other languages
German (de)
English (en)
French (fr)
Inventor
Ian c/o Imageware Res and Dev Inc Jaffray
John F. c/o Imageware Res and Dev Inc Bronskill
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Softimage Inc
Original Assignee
IMAGEWARE RESEARCH AND DEVELOPMENT Inc
Imageware Res and Dev Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by IMAGEWARE RESEARCH AND DEVELOPMENT Inc, Imageware Res and Dev Inc filed Critical IMAGEWARE RESEARCH AND DEVELOPMENT Inc
Publication of EP0703698A1 publication Critical patent/EP0703698A1/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation
    • G06T11/001—Texturing; Colouring; Generation of texture or colour
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
    • H04N1/40—Picture signal circuits
    • H04N1/40093—Modification of content of picture, e.g. retouching

Definitions

  • This invention relates to both a method and an apparatus for transforming pictures or images. More particularly, it relates to a method or apparatus for effecting a transformation of a digitized signal of an image to achieve a painted appearance, including an airbrush effect.
  • the present invention provides an apparatus and method, capable of applying an air brush appearance to a digitized signal of an image.
  • an apparatus for adding to a digitized signal of an image an airbrush effect, the apparatus comprising: means for generating, for each pixel, first and second random numbers corresponding to the first and second coordinates for that pixel; summation means for adding, for each pixel, two corresponding, random numbers to the two corresponding coordinates, to produce output coordinates, whereby the pixels of the original image are scattered randomly in an output image in dependence upon the random numbers generated; and a rank value filter connected to the summation means to receive the output coordinates therefrom and for performing a rank value operation, and having an output for a transformed image.
  • the present invention also encompasses a system or apparatus for incorporating two or more effects into a digitized signal of an image.
  • the apparatus further includes a conditioning unit for generating a conditioning signal, and also an image composition unit.
  • the image composition unit receives the outputs from the selected apparatus and also the output from the conditioning unit.
  • the composition unit then composes an output image by selective combination of the outputs of the various apparatus, in dependence upon the conditioning signal from the conditioning unit.
  • the present invention also provides methods corresponding to the apparatus.
  • each pixel is denoted by P(x,y), where x and y are the coordinates for that particular pixel.
  • P denotes the intensity of the pixel.
  • One conventional use of applying a gain to the pixels is to compensate for an image which has a predominance of low intensity pixels, i.e. the image has an overall dark appearance. If one draws a histogram of the frequency of occurrence against intensity, one gains an impression of the overall impression of the picture. If all the pixels are clustered towards the left hand end of the scale, i.e. indicating uniformly low intensity, then one can apply a certain gain to all the pixels to expand the range of intensity or grade levels to cover the entire range. Similarly, an excessively bright image will show a histogram with all the pixels clustered towards the upper end of the grade level or intensity scale. This can simply be modified by applying a gain which is less than unity, to reduce the value of the intensity.
  • Image filtering is another standard technique which is employed by the present invention in combination with other standard techniques.
  • a mean filter or blur replaces the intensity of each pixel by an intensity derived by averaging or taking arithmetic mean value of the intensity of that pixel and its neighbours. This operation is repeated for each pixel in the image.
  • a 3x3 window blur would take the values of nine pixels in a square and then use this average value as the intensity for the centre pixel of that window.
  • the inventors have realized that a variety of interesting and visually pleasing effects can be achieved by, in effect, deliberately introducing controlled distortion or noise. This gives a desired visual effect in the final image.
  • the invention makes use of four different digital classes, namely: neighbourhood operations; point transformation operations; geometrical transformation; and colour space conversion.
  • a neighbourhood operation is the modification of pixel values in a digitized image based on the value of the pixel itself and the value of nearby pixels in a pre-defined neighbourhood or window.
  • a neighbourhood transformation By performing a neighbourhood transformation on every pixel on an image, one can realize a number of different image filtering operations.
  • This is given the example of simply taking the arithmetic mean to achieve a blurring effect.
  • This is a particular example of a two-dimensional convolution (sometimes referred to as a finite impulse response filter), which simply replaces a pixel value under consideration with a weighted average of the pixel and its neighbours.
  • the particular example given above took the same value for all the pixels in the window or neighbourhood, to give a low-pass filter which blurs the image.
  • Different weights can be given to the pixels to achieve a high-pass filter which sharpens an image or a band-pass filter which enhances or suppresses certain details in an image.
  • P 2 (x,y) 4P 1 (x,y)-P 1 (x-1,y)-P 1 (x+1,y)-P 1 (x,y-1)-P 1 (x,y+1) for all x,y.
  • L in a rectangle is used to denote a Laplacian filter.
  • Another neighbourhood operation that is commonly used is a rank value filter. All the pixels in the selected neighbourhood are ordered or ranked from smallest to largest in intensity. The centre pixel in the neighbourhood is then replaced with the pixel value that has a specified rank. A median rank filter replaces the centre pixel with the pixel value that represents the middle or median rank. A maximum filter replaces the centre pixel with the maximum value in the neighbourhood, and a minimum filter operates accordingly.
  • the maximum and minimum rank filters fall into a special sub-class called morphological, which have powerful geometric properties.
  • a maximum filter is often referred to as a dilation filter, as everything expands or swells; a minimum filter is often referred to as an erosion filter, as everything shrinks.
  • RVF rank value filter
  • Neighbourhood operations can also be used to implement edge detectors.
  • An edge detector is one that outputs a high value when there is a sharp change in image intensity and outputs a low value in areas of constant intensity.
  • the output of an edge detector or edge map is useful for emphasizing or de-emphasizing the edge content in an image.
  • Various techniques have been used which depend upon edge maps derived from edge detection. In other words, the filter neighbourhood size and shape changes based on the edge magnitude and direction. This enables a variety of effects to be achieved, that are totally driven by the image content.
  • the contrast stretch outlined above is an example of a point transformation, which involves mapping a single pixel value to another, independently of other pixel values.
  • Another example of point operation is thresholding.
  • pixels that exceed a pre-defined intensity threshold are mapped to a particular value, and those that for below the threshold are mapped to another value.
  • This operation can effectively be used to divide an image into two components, often to separate a foreground object from its background.
  • the process can be generalized to multiple thresholds.
  • Such thresholds can be used to effect a pseudo-colouring of the picture, which is carried out by assigning individual colours to pre-defined intensity ranges.
  • This point transformation operation can enhance perception of certain details in an image. Since point transformations amount to a simple re-mapping of a pixel value, they can be realized with a look-up table (LUT) operation. LUT processors operating in real time are available from several companies.
  • Another type of image transformation is one that re-maps the locations of pixels in an image. An example of this would be to rotate an image through a given angle.
  • the present invention uses several novel geometrical image manipulations which are called perturbation effects, since location of a pixel is perturbed in some manner. By adding random noise to each pixel, one can achieve an airbrush or splatter paint effect, depending on the amplitude of noise added.
  • a final category of image manipulation that is used by the present invention is colour space conversion.
  • Most colour video images reside in the RGB (red, green, blue) colour space, due to the limitation of phosphor colours.
  • colour image processing is most conveniently carried out in the HSI (hue, saturation, intensity) colour space where the colour of a pixel may be decoupled from its intensity.
  • HSI hue, saturation, intensity
  • RGB to HSI and HSI to RGB conversions are commonly used in operations by the present invention.
  • CMYK cyan, magenta, yellow, key
  • Figure 1 shows an apparatus for effectively imparting a brush stroke texture to an image, the apparatus in Figure 1 being generally denoted by the reference 1.
  • the apparatus 1 includes an input 2 for the image, which is the input to a rank value filter 4.
  • the rank value filter 4 is in turn connected to a Laplacian filter 6 and then a variable gain unit 8.
  • the gain unit 8 has its output connected through an addition unit 10 to an output 12. Another input of the addition unit 10 is taken directly from the output of the rank value filter 4 through a bypass line as indicated at 14.
  • a kernel or window size and shape is selected for the rank value filter 4 and this determines the brush stroke size and shape.
  • the gain, G set by the gain unit 8, sets the stroke boldness. If G is set to zero, the stroke will be muted. However as the gain G increases, the stroke prominence increases.
  • the rank value filter 4 can have a square kernel with each dimension of the kernel varying from 1 - 15 pixels (with a median rank value).
  • the gain unit 8 can provide a gain in the range 0 - 3. Zero gain gives a muted brush stroke, whereas a gain of 3 gives a bold brush stroke affect.
  • the size of the kernel affects the brush stroke size and imparted.
  • a more particularly preferred set of parameters would be a kernel size of 7 pixels square and a gain of 1.5.
  • kernel shapes could be used, for example square, rectangular, diagonal, cross and circular, depending upon the type of brush stroke required and the direction required for the brush stroke.
  • the rank value filter 4 removes or smooths details from the image that are smaller than the filter kernel extent, hence it is the kernel size that determines the effective brush stroke size. This local smoothing action tends to leave an imprint of the size and shape of the rank value filter kernel in the ares of the image where detail has been removed. If the kernel shape and size are chosen such that it is the shape and size of the desired brush stroke, the rank value filter output image will appear to have muted brush strokes imparted on it.
  • a Laplacian filter is often employed to emphasize the image detail. Here, the Laplacian filter is employed to emphasize the boundaries of the imparted brush strokes, and depending upon the gain used, the brush stroke can range from muted to bold as the gain is increased.
  • FIG. 2 there is shown an apparatus generally indicated by the reference 20, which again has an input 22 and an output 24 which are connected through an addition or summation unit 26.
  • the input 22 is additionally connected through an edge magnitude detector unit 28 and a variable gain unit 30, whose output is connected to another input of the addition unit 26.
  • the gain unit 30 can be adjusted to provide either a positive or negative sign to the gain.
  • the effect of the units 28, 30 is to add the detected edges to the output image. If a positive sign is set by the gain unit 30, then the edges will be outlined in white, whereas if the unit 30 provides a negative sign then the edges will be outlined in black. The intensity of the outlining depends upon the gain set by the unit 30.
  • this second apparatus would have an edge magnitude detecting unit 28 which is a morphological edge detector (as disclosed in J. Serra, "Image Analysis Mathematical Morphology", Academic Press, New York, 1983).
  • This edge detector has a square kernel with each side of the kernel having from 1 - 5 pixels, more preferably 3 pixels.
  • the gain unit 30 can have a gain that varies in the range of 1 - 5 and preferably a gain of 3.5.
  • the size of the kernel and the edge detector is directly proportional to the edge thickness in the pixels.
  • edge detectors that could equally be used as the edge magnitude detector number 28 are the Sobel Edge Detector, the Compass Gradient Edge Detector, the Laplacian Edge Detector, the Roberts Edge Detector, the Kirsch Operator, the Difference of Gaussians Edge Detector. It should be noted that a variety of other image edge enhancement filters could be used.
  • the edge magnitude detector unit 28 creates an image in which each pixel in image is proportional to the magnitude of any intensity changes near that pixel. Thus, areas where intensity changes abruptly have a high output in the edge detection image, and areas with little change in intensity have a low output in the edge detection image.
  • This method strengthens the edge content of an image by adding or subtracting edges that have first been multiplied by a variable gain factor to or from the original input image. Adding the gain multiplied edges tends to make regions of the input image with high edge content to appear white, while subtracting the gain multiplied edges makes those regions appear black. Thus, the overall effect of this technique is to make areas in the input image with a high edge content become outlined in white or black.
  • Figure 3 shows an apparatus generally denoted by the reference 32 which includes an input 34 and an output 36, for the input image denoted by P i (x,y) and P o (x,y) respectively.
  • the processing is indicated within the box 38.
  • P o (x,y) P i (x+Gn 1 (x,y),y + Gn 2 (x,y)), for all x,y
  • n1(x,y) and n2(x,y) are random numbers generated for each input image pixel
  • G is a constant gain value.
  • each pixel given by the coordinates x,y one generates two random numbers n1(x,y) and n2(x,y). Each of these random numbers is multiplied by a gain factor G and then added to the respective coordinate x or y.
  • G gain factor
  • each of the output coordinates for x and y is the same as the input coordinate, plus the random number multiplied by the preset gain.
  • the effect of this is to scatter the pixels across the image, the degree of displacement of the pixels from their original positions being dependent upon the gain set. This gives an air brush effect with variable coarseness, the degree of coarseness being determined by the gain set.
  • a preferred random numbered generator is one which produces random numbers with a uniform probability density function in the range from 0 to 1. This is then preferably combined with a gain of 2 to give a mild splattering dislocation of the pixels.
  • a gain of, for example, 20 gives a very dislocated and hazy splattering of a pixel, while gains of greater than 20 tend to produce images that are unrecognizable.
  • FIG. 4 there is shown an apparatus for providing a chrome surface effect.
  • the apparatus is generally denoted by the reference 40.
  • the apparatus is shown as a single unit having an input 42 for an image, P i , to be processed and a second input 44 for an image, P R , that is to be reflected into the output image.
  • An output is indicated at 46.
  • P o (x,y) P R (X T ,Y T ) for all x,y
  • a, b are constants setting the surface smoothness
  • x m and y m represent the maximum extent of the digitized input images in the x and y directions respectively, i.e. the number of pixels in the two directions.
  • the process here is reflecting the image, P R , in the input image, P i , and thus is treating the input image as a reflective or mirrored surface.
  • the intensity of each pixel in the input image, P i is treated as the height above an arbitrary flat surface, so as to give a three dimensional effect, two dimensions being the x and y coordinates and the third dimension being the pixel intensity.
  • the method starts by converting the input image, P i , into a three dimensional surface. It then assumes that this is reflective and effectively takes the reflection of the image, P R , in this reflective surface. In order to be able to "see” the shape of a complex reflective surface, one has to have some image that is reflected in it. It is for this reason that the image P R is provided.
  • the image P R can be any suitable image, and can be selected to give a desired appearance.
  • the input image, P i is simply a flat surface, i.e. a conventional plain mirror, then one would obtain a pure reflection of the image to be reflected, P R .
  • the input image P i is a complex shape, e.g. a person's head, then the reflective surface is extremely complex and, resulting in considerable distortion of the image to be reflected, P R , so that this is often unrecognizable. Even if the reflected image P R becomes totally distorted and unrecognizable the output image still retains the shape or appearance of the input image P i , but with a simulated, reflective or chrome finish.
  • the apparatus 50 has an input 52 for an input image which is divided into two branches, one branch 53 connected directly to a combination unit 58 and another branch 54 connected to a contrast stretch unit 56.
  • the output of the contrast stretch unit 56 is also connected to an input of the combination unit 58.
  • the combination unit 58 has an output 59.
  • the unit 56 performs a contrast stretch operation which is given by the following equation: and MAX-VAL is the maximum allowable pixel value in the input image; INTENSITY1, INTENSITY2 are selected image grey levels with INTENSITY1 > INTENSITY2.
  • the function given by the above equation essentially sets the output, P2 (x,y), by three separate calculations, depending upon the value of the input signal, P1(x,y). If P1 is less than INTENSITY2, then the output P2 is set to zero. If P1 is between INTENSITY2 and INTENSITY1, then P2 is determined by the equation above which essentially gives a straight line slope from zero to the maximum value as P1 increases from INTENSITY2 to INTENSITY1. Where P1 is greater than INTENSITY1, then the output is set to the maximum value.
  • the effect of this is to stretch a middle range of grey levels, and eliminate the upper and lower grey levels from the input signal by setting them to zero or the maximum value respectively. If one considered a histogram of the distribution of the pixel intensities against the grey level or intensity, one would find that the middle portion of the histogram had effectively been taken and stretched to cover the whole scale, whilst the outer portions of the original histogram had effectively moved to the very edges.
  • the combining function performed by the combination unit 58 can be given by either one of the following equations:
  • the first of these equations is a simple summation, and will effectively give an increase in the overall intensity.
  • the second of these equations represents an averaging effect.
  • the overall effect of this technique is to add highlights to an image.
  • the values selected for INTENSITY1 and INTENSITY2 set the highlight brightness and extent.
  • INTENSITY2 and INTENSITY1 could be chosen as the sixtieth percentile grey level in the input image and the ninety-fifth percentile grey level in the input image respectively.
  • This percentile selection adds robustness to a varying lighting condition. This effectively adds or averages the pixel intensities between the sixtieth and ninety-fifth intensity percentiles to the input image. This range of intensities between these two percentiles is deemed to be the highlights of the input image.
  • the highlights are averaged with the input image, the highlights are incorporated into the image in the locations that they are present in the original input image; however, in areas of image where there are no highlights present, the addition of highlights has no effect.
  • the areas with highlights are still highlighted, but to a slightly lesser extent, whereas the areas with no highlights are effectively decreased in intensity. This has the effect of making the highlights more pronounced.
  • Averaging the highlights into the image makes the output image appear as if the highlights were added using chalk.
  • the apparatus here denoted 60, has an input 62 connected to first and second mean filters 63, 64.
  • the output of the mean filters are connected to positive and negative inputs of a summation unit 66, which has an output 68 forming the output of the apparatus.
  • the first mean filter 63 has a kernel m x n
  • the second mean filter has a kernel u x v.
  • the kernel of the first mean filter 63 is greater than that of the second mean filter 64; in other words, m is greater than u and n is greater than v.
  • the output at 68 is given by the following equation:
  • the effect of this arrangement is, for each pixel, to first take a mean within a first kernel of all the pixels in that kernel, and then subtract a mean signal derived from the second, smaller kernel, to arrive at an output signal.
  • Each mean filter 63, 64 performs a low-pass function.
  • the cut-off frequency of each mean filter is determined by the size of the kernel, so that the filter with a smaller kernel has a higher cut off frequency.
  • edge information occupies the higher frequency regions of an image, i.e. sharp transitions.
  • image noise also tends to reside at the higher frequencies.
  • a band-pass filter one can pass some of the high frequencies through to extract the image edges for forming a line drawing, but simultaneously attenuate the highest frequencies that contain noise and make for a dirtier or noisier line drawing image.
  • Figure 7 shows an apparatus for modifying an image so that it appears to be painted in a water colour style.
  • rounded blobby features reminiscent of, or simulating, paint dabs are added to the image.
  • the apparatus 70 of Figure 7 has an input 72 connected to an input of a first rank value filter 74, which in turn has an output connected to a second rank value filter 76.
  • the output of the second rank value filter 76 is connected, as in the first arrangement of Figure 1, through a Laplacian unit 78 and a gain unit 80 to a summation unit 82. There is also a bypass line 84 providing a direct connection from the output of the filter 76 to the summation unit 82.
  • the summation unit 82 sums its two inputs and forms an output 86.
  • the two rank value filters 74, 76 have identical kernel size and shape, but the rank value for each filter is chosen differently, in accordance with the following method.
  • a rank value of 1 with respect to a kernel correspond to the minimum pixel value in the kernel and a rank value of N correspond to the maximum pixel values in the kernel.
  • a value of p such that: 1 ⁇ p ⁇ N
  • the rank for the filters 74, 76 are selected as: RVF Filter 74: p RVF Filter 76: (N + 1) - p
  • p is chosen arbitrarily and the sum of the two ranks for the two rank value filters is equal to the sum of the maximum and minimum rank value in the kernel.
  • the rank for each filter will be similar. The bright areas of the image do not then move relative to the dark areas of the image.
  • the first rank value filter will have the low rank p, whilst the second rank value filter 76 will have a relatively high rank. This has the effect of the dark areas of the image expanding more into the light regions.
  • the light regions of the image expand more into the dark regions.
  • the combination of the two rank value filters produces the rounded blobby areas.
  • the units 78-84 accentuates the paint dabs.
  • a low gain e.g. close to zero, produces a muted blob, whilst a higher gain produces a stronger dab.
  • components 78-84 correspond to the arrangement shown in Figure 1.
  • the first rank value filter 74 is a local minimum filter or morphological erosion operator, i.e. it causes bright areas of the image to contract and dark areas to expand
  • the second rank value filter 76 is then a local maximum filter or dilation operator, i.e. bright areas of the image expand while dark areas contract.
  • the combination of the two filters operating as erosion and dilation operators performs an operation referred to as a morphological opening.
  • the net effect of an opening is that local peaks in the image smaller than the kernel extent are smoothed from the image and the dark areas of the image seep into the bright areas, since the dilation does not quite counter-act the initial erosion.
  • the combination of this local peak smoothing and dark regions swelling produces round blobby areas in the image reminiscent of water colour paint dabs.
  • the roles of the two rank value filters are reversed.
  • the first rank value filter 74 becomes a maximum filter, whilst the second rank value filter 76 becomes a local minimum filter.
  • the combination of the two filters working in series then performs a morphological closing.
  • the net effect of such a closing is the local valleys in the image, i.e. dark areas which are smaller than the kernel extent, are filled in and the bright areas of the image seep into the dark areas.
  • the erosion does not quite counteract the initial dilation.
  • valley filling and light region swelling produces blobby areas reminiscent of water colour dabs.
  • the role of the Laplacian filter 78 and gain unit 80 is to strengthen the paint dab boundaries. The higher the gain the more pronounced the boundary.
  • the apparatus 90 has an input 92 connected to first and second processes indicated at 94, 96 and to a conditioning unit 98. The outputs of these three units 94, 96 and 98 are connected to an image composition unit 100 which produces an output 102.
  • the processes 94, 96 can be any one of the processes in accordance with the present invention, e.g. those described in relation to the preceding figures. This apparatus enables them to be combined in a variety of ways.
  • the conditioning unit 98 provides a switching function to combine the two modified images produced from the processes 94, 96 as desired.
  • this function provides that the respective weights given to the two processes A, B, depends upon the intensity of the conditioning signal, C, for that particular pixel.
  • conditioning unit 98 useful conditioning functions for the conditioning unit 98 are: no conditioning performed; edge magnitude detection; and contrast stretching. Other conditioning techniques are possible. Thus, one can detect different areas of an image in relation to colour and/or intensity or other factors. Then, these different areas can be subjected to different processes. Also, whilst just two processes 94, 96 are shown, it will be realized that this basic arrangement can be generalized to any number of processes.
  • Another possibility is to combine images dependent upon the brightness, i.e. in the bright areas one processing technique is used, whereas in the dark areas another technique is used.
  • the input image itself may serve as the switching function.
  • an edge magnitude detector could be employed to create image C. This has the effect of having image A dominate the output image and areas of high edge intensity and image B in regions of low edge intensity.
  • the input image could have its intensity profile modified in some way such as a contrast stretch in order to modify the switching function.
  • the conditioning unit 98 has an input 104 which is connected to the inputs of a rank value filter 106 and a mean filter 108.
  • the outputs of these two filters 106, 108 are connected to a combination unit 110 which has positive and negative inputs for the two filters, 106, 108 respectively.
  • the output of the unit 110 is connected to the threshold unit 112, and in turn to an output 114.
  • the rank value filter 106 has a rank value of 25, i.e. a median value.
  • the thresholding unit 112 provides thresholding process where every pixel intensity greater than the threshold t is mapped to MAX VAL. Any pixel having an intensity less than t is mapped to zero.
  • t is set equal to 1.
  • the output of 114 will be set equal to MAX VAL, where the local median value is greater than or equal to the local mean value. On the other hand, where the median value is less than the mean value, the output 114 will be zero.
  • the two filters 106, 108 preferably have a kernel size of 7 x 7.
  • FIG. 10 shows a block diagram for a real-time digital video effect process, indicated by the reference 120.
  • the process of 120 has an analog to digital converter 122 with an input for a video signal. This produces two outputs, 123, 124 for the RGB and HSI colour spaces.
  • a switch 126 enables either or both of these outputs 123, 124 to be connected through to two separate branches 128 and 130.
  • the first branch 128 there is an rank value filter 132, connected to a convolution filter 134, and then in turn to a lookup table 136.
  • the second branch 130 there is an edge detection unit 138, another lookup table 140 and an arithmetic logic unit 142.
  • the various components 132-142 would be mounted in a common housing and connected, as indicated by terminals 146, to one or more digital crosspoint switches. These digital crosspoint switches would enable the components 132-142 to be connected in a variety of patterns.
  • the input switch 126 and output 148 are similarly provided with terminals 146 to enable them to be connected by the digital crosspoint switches.
  • arrows 150 indicate, schematically, the digital crosspoint switch or switches and their effective connections.
  • Such functional units include: variable gain units; Laplacian units; random number generators; contrast stretch units; and mean filters.
  • the input signal passes through the first branch 128 where the signal is given a brush stroke effect by the rank value filter 132 and then sharpened in the convolution filter 138 prior to a contrast stretch operation by the lookup table 136.
  • the convolution filter 138 detects edges, and the magnitude of the edges and then normalized by the lookup table 140.
  • the arithmetical logic unit 142 subtracts the output of the two lookup tables 136, 140, so as to subtract the normalized edges from the image from the first branch 128.
  • the edges in the resulting image will now have dark outline highlights.
  • the output 148 is then connected by a switch 152 to the RGB or HSI input of a digital to analog converter 154, and then to a final output 156.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Image Processing (AREA)
EP95117080A 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect Withdrawn EP0703698A1 (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US387049 1989-07-31
US07/387,049 US5063448A (en) 1989-07-31 1989-07-31 Apparatus and method for transforming a digitized signal of an image
EP90911858A EP0485459B1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
EP90911858.0 Division 1990-07-31

Publications (1)

Publication Number Publication Date
EP0703698A1 true EP0703698A1 (en) 1996-03-27

Family

ID=23528236

Family Applications (3)

Application Number Title Priority Date Filing Date
EP90911858A Expired - Lifetime EP0485459B1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image
EP95117078A Withdrawn EP0707409A1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image into a reflective surface
EP95117080A Withdrawn EP0703698A1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect

Family Applications Before (2)

Application Number Title Priority Date Filing Date
EP90911858A Expired - Lifetime EP0485459B1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image
EP95117078A Withdrawn EP0707409A1 (en) 1989-07-31 1990-07-31 Apparatus and method for transforming a digitized signal of an image into a reflective surface

Country Status (7)

Country Link
US (2) US5063448A (ja)
EP (3) EP0485459B1 (ja)
JP (1) JP3009726B2 (ja)
AU (1) AU6064390A (ja)
CA (1) CA1333636C (ja)
DE (1) DE69029136T2 (ja)
WO (1) WO1991002425A1 (ja)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1400922A1 (en) * 2002-09-20 2004-03-24 Nitto Denko Corporation Print inspection method and apparatus
WO2009097279A1 (en) * 2008-01-28 2009-08-06 Vistaprint Technologies Limited Creating images for displaying or printing on low-contrast background

Families Citing this family (43)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5245432A (en) * 1989-07-31 1993-09-14 Imageware Research And Development Inc. Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect
JP2774627B2 (ja) * 1989-12-28 1998-07-09 株式会社日立製作所 画像表示方法及びその装置
US5428775A (en) * 1990-05-24 1995-06-27 Apple Computer, Inc. Apparatus for providing data dependent write operations
US5268860A (en) * 1990-09-17 1993-12-07 Victor Company Of Japan, Ltd. Image enhancement system
JP2940317B2 (ja) * 1992-06-24 1999-08-25 三菱電機株式会社 画像処理装置
US5519618A (en) * 1993-08-02 1996-05-21 Massachusetts Institute Of Technology Airport surface safety logic
US5374932A (en) * 1993-08-02 1994-12-20 Massachusetts Institute Of Technology Airport surface surveillance system
US5563962A (en) * 1994-03-08 1996-10-08 The University Of Connecticut Two dimensional digital hysteresis filter for smoothing digital images
US5592571A (en) * 1994-03-08 1997-01-07 The University Of Connecticut Digital pixel-accurate intensity processing method for image information enhancement
WO1995024694A1 (en) * 1994-03-08 1995-09-14 The University Of Connecticut Digital pixel-accurate intensity processing method for image information enhancement
US5621868A (en) * 1994-04-15 1997-04-15 Sony Corporation Generating imitation custom artwork by simulating brush strokes and enhancing edges
DE69518578T2 (de) * 1994-05-18 2001-04-26 Sharp Kk Kartenartige Kamera mit Bildverarbeitungsfunktion
US5847712A (en) * 1995-01-03 1998-12-08 University Of Washington Method and system for generating graphic illustrations according to a stroke texture and a tone
US7616198B2 (en) * 1998-02-20 2009-11-10 Mental Images Gmbh System and computer-implemented method for modeling the three-dimensional shape of an object by shading of a two-dimensional image of the object
US6011536A (en) * 1998-04-17 2000-01-04 New York University Method and system for generating an image having a hand-painted appearance
DE69927239T2 (de) 1998-09-15 2006-07-13 Phase One A/S System und verfahren zur verarbeitung von bildern
US6788812B1 (en) 1999-06-18 2004-09-07 Eastman Kodak Company Techniques for selective enhancement of a digital image
US6522427B1 (en) * 1999-09-21 2003-02-18 Seiko Epson Corporation Color table manipulations for contour reduction
US7023576B1 (en) * 2000-05-09 2006-04-04 Phase One A/S Method and an apparatus for elimination of color Moiré
US6804418B1 (en) * 2000-11-03 2004-10-12 Eastman Kodak Company Petite size image processing engine
US7392287B2 (en) 2001-03-27 2008-06-24 Hemisphere Ii Investment Lp Method and apparatus for sharing information using a handheld device
US7103234B2 (en) * 2001-03-30 2006-09-05 Nec Laboratories America, Inc. Method for blind cross-spectral image registration
US20040233196A1 (en) * 2001-11-13 2004-11-25 Hertzmann Aaron P Logic arrangements storage mediums, and methods for generating digital images using brush strokes
CN1303570C (zh) 2002-02-12 2007-03-07 松下电器产业株式会社 图象处理装置和图象处理方法
US20050153356A1 (en) * 2002-09-09 2005-07-14 Olympus Corporation Image processing method for biochemical test
US7680342B2 (en) * 2004-08-16 2010-03-16 Fotonation Vision Limited Indoor/outdoor classification in digital images
US7606417B2 (en) 2004-08-16 2009-10-20 Fotonation Vision Limited Foreground/background segmentation in digital images with differential exposure calculations
US7761363B2 (en) * 2003-10-08 2010-07-20 Fx Alliance, Llc Internal trade requirement order management and execution system
US7692696B2 (en) * 2005-12-27 2010-04-06 Fotonation Vision Limited Digital image acquisition system with portrait mode
US7469071B2 (en) 2006-02-14 2008-12-23 Fotonation Vision Limited Image blurring
IES20060558A2 (en) * 2006-02-14 2006-11-01 Fotonation Vision Ltd Image blurring
IES20060564A2 (en) 2006-05-03 2006-11-01 Fotonation Vision Ltd Improved foreground / background separation
JP4635975B2 (ja) * 2006-07-20 2011-02-23 パナソニック株式会社 画像処理装置及び画像処理方法
US8214497B2 (en) 2007-01-24 2012-07-03 Mcafee, Inc. Multi-dimensional reputation scoring
US8763114B2 (en) * 2007-01-24 2014-06-24 Mcafee, Inc. Detecting image spam
US8184143B2 (en) * 2008-06-27 2012-05-22 Sony Mobile Communications Ab Simulated reflective display
TW201118791A (en) * 2009-11-27 2011-06-01 Inst Information Industry System and method for obtaining camera parameters from a plurality of images, and computer program products thereof
JP2010166597A (ja) * 2010-03-10 2010-07-29 Panasonic Corp 画像処理装置及び画像処理方法
JP6004757B2 (ja) * 2012-06-07 2016-10-12 キヤノン株式会社 画像処理装置及び画像処理方法
JP2015011130A (ja) * 2013-06-27 2015-01-19 株式会社横須賀テレコムリサーチパーク 画像処理装置、電気泳動表示装置、画像処理方法、及びプログラム
CN105185351B (zh) * 2015-10-13 2017-07-28 深圳市华星光电技术有限公司 提升oled显示面板对比度的方法及系统
TWI677233B (zh) * 2018-08-02 2019-11-11 瑞昱半導體股份有限公司 決定濾波器係數的方法
CN110675343B (zh) * 2019-09-24 2023-02-24 西安科技大学 一种井下输煤图像的图像增强方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE3546135A1 (de) * 1984-12-28 1986-08-14 Canon K.K., Tokio/Tokyo Verfahren und einrichtung zur bildverarbeitung
EP0216536A2 (en) * 1985-08-29 1987-04-01 Canon Kabushiki Kaisha Image processing apparatus
EP0270259A2 (en) * 1986-12-04 1988-06-08 Quantel Limited Improvements relating to video signal processing systems

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6031389B2 (ja) * 1978-04-05 1985-07-22 日本電気株式会社 テレビジョン映像信号特殊効果装置
DE3620261A1 (de) * 1986-06-16 1987-12-23 Ruediger Dr Brennecke Verfahren zur ueberlagerung unterschiedlicher bilder
US5124784A (en) * 1986-07-25 1992-06-23 Canon Kabushiki Kaisha Video signal processing apparatus
US4782388A (en) * 1986-10-24 1988-11-01 The Grass Valley Group, Inc. Method and apparatus for providing video mosaic effects

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE3546135A1 (de) * 1984-12-28 1986-08-14 Canon K.K., Tokio/Tokyo Verfahren und einrichtung zur bildverarbeitung
EP0216536A2 (en) * 1985-08-29 1987-04-01 Canon Kabushiki Kaisha Image processing apparatus
EP0270259A2 (en) * 1986-12-04 1988-06-08 Quantel Limited Improvements relating to video signal processing systems

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
J.-H. LEE ET AL.: "A FAST ALGORITHM FOR TWO-DIMENSIONAL WILCOXON FILTERING", 1987 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS, vol. 1 OF 3, 4 May 1987 (1987-05-04) - 7 May 1987 (1987-05-07), PHILADELPHIA, PA, pages 268 - 271 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1400922A1 (en) * 2002-09-20 2004-03-24 Nitto Denko Corporation Print inspection method and apparatus
US7260244B2 (en) 2002-09-20 2007-08-21 Nitto Denko Corporation Print inspection method and print inspection apparatus
WO2009097279A1 (en) * 2008-01-28 2009-08-06 Vistaprint Technologies Limited Creating images for displaying or printing on low-contrast background
EP2372658A1 (en) * 2008-01-28 2011-10-05 VistaPrint Technologies Limited Creating images for displaying or printing on low-contrast background
CN101925927B (zh) * 2008-01-28 2013-05-22 威仕达品特技术有限公司 创建用于在低对比背景上显示或打印的图像
US8810603B2 (en) 2008-01-28 2014-08-19 Vistaprint Schweiz Gmbh Creating images for displaying or printing on low-contrast background

Also Published As

Publication number Publication date
JP3009726B2 (ja) 2000-02-14
DE69029136T2 (de) 1997-05-22
WO1991002425A1 (en) 1991-02-21
US5063448A (en) 1991-11-05
EP0485459A1 (en) 1992-05-20
US5325200A (en) 1994-06-28
EP0707409A1 (en) 1996-04-17
JPH04507314A (ja) 1992-12-17
AU6064390A (en) 1991-03-11
CA1333636C (en) 1994-12-20
EP0485459B1 (en) 1996-11-13
DE69029136D1 (de) 1996-12-19

Similar Documents

Publication Publication Date Title
EP0485459B1 (en) Apparatus and method for transforming a digitized signal of an image
US5245432A (en) Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect
US9626751B2 (en) Method and system for analog/digital image simplification and stylization
KR100369909B1 (ko) 원영상을수정하는장치및방법
US5247583A (en) Image segmentation method and apparatus therefor
US6088487A (en) Apparatus and method for changing a video image to a drawing-style image
CA2519627A1 (en) Selective enhancement of digital images
Pinoli et al. Logarithmic adaptive neighborhood image processing (LANIP): Introduction, connections to human brightness perception, and application issues
Wu et al. Performance evaluation of some noise reduction methods
CA1340551C (en) Apparatus and method for transforming a digitized signal of an image to incorporate an airbrush effect
US20040130554A1 (en) Application of visual effects to a region of interest within an image
Bangham et al. The Art of Scale-Space.
Mitchell The antialiasing problem in ray tracing
du Buf et al. Painterly rendering using human vision
Lehar et al. Image processing system for enhancement and deblurring of photographs
JPH10187964A (ja) 画像フィルタリング方法および画像フィルタリング装置並びに画像の輪郭強調処理装置
Boaventura et al. Border detection in digital images: An approach by fuzzy numbers
Kassab Image enhancement methods and implementation in MATLAB
Waltz Image processing operations in color space using finite-state machines
Tang Application of non-parametric texture synthesis to image inpainting
Aman et al. Image Processing by Bilateral Filtering Method
Brigner The problem of the missing wavelet
Rieke Adriati Wijayanti et al. WORKSHOP IMAGE PROCESSING
Bahal et al. De-Noising of Color Image by Removing Random Impulse Noise Using MATLAB
Lenka et al. Fusion of and bilateral histogram with Retinex Algorithm for Enhancement

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 19951124

AC Divisional application: reference to earlier application

Ref document number: 485459

Country of ref document: EP

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): DE FR GB IT

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: SOFTIMAGE, INC.

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION HAS BEEN WITHDRAWN

18W Application withdrawn

Withdrawal date: 19961007